OBJECTIVE To compare the correlation and practicability of single-stage vs. graded treadmill protocols in the assessment of the absolute claudication distance (ACD). METHODS Randomized open study. METHODS In 52 patients with peripheral arterial occlusive disease, the ACD on treadmill at 3 km/h and 12% grade (constant-load test = C-test) ranged form 50 to 400 m. The C-test and the graded-exercise test (walking on the treadmill at 3 km/h and 0% gradient for 3 min, with subsequent increase in gradient of 3.5% every 3 min = G-test) were carried out at random on the same day under standardized conditions. RESULTS The ACD was higher in the G-test than in the C-test (360.4 +/- 208.8m vs. 166.5 +/- 93.6m; p < 0.001). The coefficients of variation were very similar (57.9% and 56.2%, respectively). In the subgroup of patients with an ACD of between 100 m and 150 m, a large difference was found both for the coefficient of variation (58.6% G-test, 9.6% C-test) and for the standard deviation (339.8 +/- 199.0m and 133.1 +/- 12.8m, respectively). CONCLUSIONS For the assessment of the ACD in patients with severe claudication the C-test would seem to be more suitable than the G-test.